EEG Sensor Network Based Feature Extraction for Predicting Brain Disorders
DescriptionThe research done here utilizes electroencephalogram (EEG) brain signals to identify brain disorders namely: schizophrenia, insomnia, epilepsy, autism, and alcoholism from a healthy brain. This system is based on pre-trained data accumulated from several different datasets, one for each brain disorder.
This research proposes a complex network-based feature extraction method, for predicting brain disorders in an unlabeled brain network. Analysis of EEG sensor networks produce a complex brain network that provides greater insights into the patterns of brain connectivity. This allows allowing for better aggregation of datasets. This accounted for the variations in the method of data collection. Features from the network were used for training Machine learning models like SVM, K-nearest neighbor and logistic regression were compared to which K-nearest neighbor with an accuracy of 85%.